1/*
2 * Licensed to the Apache Software Foundation (ASF) under one or more
3 * contributor license agreements.  See the NOTICE file distributed with
4 * this work for additional information regarding copyright ownership.
5 * The ASF licenses this file to You under the Apache License, Version 2.0
6 * (the "License"); you may not use this file except in compliance with
7 * the License.  You may obtain a copy of the License at
8 *
9 *      http://www.apache.org/licenses/LICENSE-2.0
10 *
11 * Unless required by applicable law or agreed to in writing, software
12 * distributed under the License is distributed on an "AS IS" BASIS,
13 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14 * See the License for the specific language governing permissions and
15 * limitations under the License.
16 */
17
18package org.apache.commons.math.random;
19
20import java.io.Serializable;
21import java.security.MessageDigest;
22import java.security.NoSuchAlgorithmException;
23import java.security.NoSuchProviderException;
24import java.security.SecureRandom;
25import java.util.Collection;
26
27import org.apache.commons.math.MathException;
28import org.apache.commons.math.distribution.BetaDistributionImpl;
29import org.apache.commons.math.distribution.BinomialDistributionImpl;
30import org.apache.commons.math.distribution.CauchyDistributionImpl;
31import org.apache.commons.math.distribution.ChiSquaredDistributionImpl;
32import org.apache.commons.math.distribution.ContinuousDistribution;
33import org.apache.commons.math.distribution.FDistributionImpl;
34import org.apache.commons.math.distribution.GammaDistributionImpl;
35import org.apache.commons.math.distribution.HypergeometricDistributionImpl;
36import org.apache.commons.math.distribution.IntegerDistribution;
37import org.apache.commons.math.distribution.PascalDistributionImpl;
38import org.apache.commons.math.distribution.TDistributionImpl;
39import org.apache.commons.math.distribution.WeibullDistributionImpl;
40import org.apache.commons.math.distribution.ZipfDistributionImpl;
41import org.apache.commons.math.exception.MathInternalError;
42import org.apache.commons.math.exception.NotStrictlyPositiveException;
43import org.apache.commons.math.exception.NumberIsTooLargeException;
44import org.apache.commons.math.exception.util.LocalizedFormats;
45import org.apache.commons.math.util.FastMath;
46import org.apache.commons.math.util.MathUtils;
47
48/**
49 * Implements the {@link RandomData} interface using a {@link RandomGenerator}
50 * instance to generate non-secure data and a {@link java.security.SecureRandom}
51 * instance to provide data for the <code>nextSecureXxx</code> methods. If no
52 * <code>RandomGenerator</code> is provided in the constructor, the default is
53 * to use a generator based on {@link java.util.Random}. To plug in a different
54 * implementation, either implement <code>RandomGenerator</code> directly or
55 * extend {@link AbstractRandomGenerator}.
56 * <p>
57 * Supports reseeding the underlying pseudo-random number generator (PRNG). The
58 * <code>SecurityProvider</code> and <code>Algorithm</code> used by the
59 * <code>SecureRandom</code> instance can also be reset.
60 * </p>
61 * <p>
62 * For details on the default PRNGs, see {@link java.util.Random} and
63 * {@link java.security.SecureRandom}.
64 * </p>
65 * <p>
66 * <strong>Usage Notes</strong>:
67 * <ul>
68 * <li>
69 * Instance variables are used to maintain <code>RandomGenerator</code> and
70 * <code>SecureRandom</code> instances used in data generation. Therefore, to
71 * generate a random sequence of values or strings, you should use just
72 * <strong>one</strong> <code>RandomDataImpl</code> instance repeatedly.</li>
73 * <li>
74 * The "secure" methods are *much* slower. These should be used only when a
75 * cryptographically secure random sequence is required. A secure random
76 * sequence is a sequence of pseudo-random values which, in addition to being
77 * well-dispersed (so no subsequence of values is an any more likely than other
78 * subsequence of the the same length), also has the additional property that
79 * knowledge of values generated up to any point in the sequence does not make
80 * it any easier to predict subsequent values.</li>
81 * <li>
82 * When a new <code>RandomDataImpl</code> is created, the underlying random
83 * number generators are <strong>not</strong> initialized. If you do not
84 * explicitly seed the default non-secure generator, it is seeded with the
85 * current time in milliseconds on first use. The same holds for the secure
86 * generator. If you provide a <code>RandomGenerator</code> to the constructor,
87 * however, this generator is not reseeded by the constructor nor is it reseeded
88 * on first use.</li>
89 * <li>
90 * The <code>reSeed</code> and <code>reSeedSecure</code> methods delegate to the
91 * corresponding methods on the underlying <code>RandomGenerator</code> and
92 * <code>SecureRandom</code> instances. Therefore, <code>reSeed(long)</code>
93 * fully resets the initial state of the non-secure random number generator (so
94 * that reseeding with a specific value always results in the same subsequent
95 * random sequence); whereas reSeedSecure(long) does <strong>not</strong>
96 * reinitialize the secure random number generator (so secure sequences started
97 * with calls to reseedSecure(long) won't be identical).</li>
98 * <li>
99 * This implementation is not synchronized.
100 * </ul>
101 * </p>
102 *
103 * @version $Revision: 1061496 $ $Date: 2011-01-20 21:32:16 +0100 (jeu. 20 janv. 2011) $
104 */
105public class RandomDataImpl implements RandomData, Serializable {
106
107    /** Serializable version identifier */
108    private static final long serialVersionUID = -626730818244969716L;
109
110    /** underlying random number generator */
111    private RandomGenerator rand = null;
112
113    /** underlying secure random number generator */
114    private SecureRandom secRand = null;
115
116    /**
117     * Construct a RandomDataImpl.
118     */
119    public RandomDataImpl() {
120    }
121
122    /**
123     * Construct a RandomDataImpl using the supplied {@link RandomGenerator} as
124     * the source of (non-secure) random data.
125     *
126     * @param rand
127     *            the source of (non-secure) random data
128     * @since 1.1
129     */
130    public RandomDataImpl(RandomGenerator rand) {
131        super();
132        this.rand = rand;
133    }
134
135    /**
136     * {@inheritDoc}
137     * <p>
138     * <strong>Algorithm Description:</strong> hex strings are generated using a
139     * 2-step process.
140     * <ol>
141     * <li>
142     * len/2+1 binary bytes are generated using the underlying Random</li>
143     * <li>
144     * Each binary byte is translated into 2 hex digits</li>
145     * </ol>
146     * </p>
147     *
148     * @param len
149     *            the desired string length.
150     * @return the random string.
151     * @throws NotStrictlyPositiveException if {@code len <= 0}.
152     */
153    public String nextHexString(int len) {
154        if (len <= 0) {
155            throw new NotStrictlyPositiveException(LocalizedFormats.LENGTH, len);
156        }
157
158        // Get a random number generator
159        RandomGenerator ran = getRan();
160
161        // Initialize output buffer
162        StringBuilder outBuffer = new StringBuilder();
163
164        // Get int(len/2)+1 random bytes
165        byte[] randomBytes = new byte[(len / 2) + 1];
166        ran.nextBytes(randomBytes);
167
168        // Convert each byte to 2 hex digits
169        for (int i = 0; i < randomBytes.length; i++) {
170            Integer c = Integer.valueOf(randomBytes[i]);
171
172            /*
173             * Add 128 to byte value to make interval 0-255 before doing hex
174             * conversion. This guarantees <= 2 hex digits from toHexString()
175             * toHexString would otherwise add 2^32 to negative arguments.
176             */
177            String hex = Integer.toHexString(c.intValue() + 128);
178
179            // Make sure we add 2 hex digits for each byte
180            if (hex.length() == 1) {
181                hex = "0" + hex;
182            }
183            outBuffer.append(hex);
184        }
185        return outBuffer.toString().substring(0, len);
186    }
187
188    /**
189     * Generate a random int value uniformly distributed between
190     * <code>lower</code> and <code>upper</code>, inclusive.
191     *
192     * @param lower
193     *            the lower bound.
194     * @param upper
195     *            the upper bound.
196     * @return the random integer.
197     * @throws NumberIsTooLargeException if {@code lower >= upper}.
198     */
199    public int nextInt(int lower, int upper) {
200        if (lower >= upper) {
201            throw new NumberIsTooLargeException(LocalizedFormats.LOWER_BOUND_NOT_BELOW_UPPER_BOUND,
202                                                lower, upper, false);
203        }
204        double r = getRan().nextDouble();
205        return (int) ((r * upper) + ((1.0 - r) * lower) + r);
206    }
207
208    /**
209     * Generate a random long value uniformly distributed between
210     * <code>lower</code> and <code>upper</code>, inclusive.
211     *
212     * @param lower
213     *            the lower bound.
214     * @param upper
215     *            the upper bound.
216     * @return the random integer.
217     * @throws NumberIsTooLargeException if {@code lower >= upper}.
218     */
219    public long nextLong(long lower, long upper) {
220        if (lower >= upper) {
221            throw new NumberIsTooLargeException(LocalizedFormats.LOWER_BOUND_NOT_BELOW_UPPER_BOUND,
222                                                lower, upper, false);
223        }
224        double r = getRan().nextDouble();
225        return (long) ((r * upper) + ((1.0 - r) * lower) + r);
226    }
227
228    /**
229     * {@inheritDoc}
230     * <p>
231     * <strong>Algorithm Description:</strong> hex strings are generated in
232     * 40-byte segments using a 3-step process.
233     * <ol>
234     * <li>
235     * 20 random bytes are generated using the underlying
236     * <code>SecureRandom</code>.</li>
237     * <li>
238     * SHA-1 hash is applied to yield a 20-byte binary digest.</li>
239     * <li>
240     * Each byte of the binary digest is converted to 2 hex digits.</li>
241     * </ol>
242     * </p>
243     *
244     * @param len
245     *            the length of the generated string
246     * @return the random string
247     * @throws NotStrictlyPositiveException if {@code len <= 0}.
248     */
249    public String nextSecureHexString(int len) {
250        if (len <= 0) {
251            throw new NotStrictlyPositiveException(LocalizedFormats.LENGTH, len);
252        }
253
254        // Get SecureRandom and setup Digest provider
255        SecureRandom secRan = getSecRan();
256        MessageDigest alg = null;
257        try {
258            alg = MessageDigest.getInstance("SHA-1");
259        } catch (NoSuchAlgorithmException ex) {
260            // this should never happen
261            throw new MathInternalError(ex);
262        }
263        alg.reset();
264
265        // Compute number of iterations required (40 bytes each)
266        int numIter = (len / 40) + 1;
267
268        StringBuilder outBuffer = new StringBuilder();
269        for (int iter = 1; iter < numIter + 1; iter++) {
270            byte[] randomBytes = new byte[40];
271            secRan.nextBytes(randomBytes);
272            alg.update(randomBytes);
273
274            // Compute hash -- will create 20-byte binary hash
275            byte hash[] = alg.digest();
276
277            // Loop over the hash, converting each byte to 2 hex digits
278            for (int i = 0; i < hash.length; i++) {
279                Integer c = Integer.valueOf(hash[i]);
280
281                /*
282                 * Add 128 to byte value to make interval 0-255 This guarantees
283                 * <= 2 hex digits from toHexString() toHexString would
284                 * otherwise add 2^32 to negative arguments
285                 */
286                String hex = Integer.toHexString(c.intValue() + 128);
287
288                // Keep strings uniform length -- guarantees 40 bytes
289                if (hex.length() == 1) {
290                    hex = "0" + hex;
291                }
292                outBuffer.append(hex);
293            }
294        }
295        return outBuffer.toString().substring(0, len);
296    }
297
298    /**
299     * Generate a random int value uniformly distributed between
300     * <code>lower</code> and <code>upper</code>, inclusive. This algorithm uses
301     * a secure random number generator.
302     *
303     * @param lower
304     *            the lower bound.
305     * @param upper
306     *            the upper bound.
307     * @return the random integer.
308     * @throws NumberIsTooLargeException if {@code lower >= upper}.
309     */
310    public int nextSecureInt(int lower, int upper) {
311        if (lower >= upper) {
312            throw new NumberIsTooLargeException(LocalizedFormats.LOWER_BOUND_NOT_BELOW_UPPER_BOUND,
313                                                lower, upper, false);
314        }
315        SecureRandom sec = getSecRan();
316        return lower + (int) (sec.nextDouble() * (upper - lower + 1));
317    }
318
319    /**
320     * Generate a random long value uniformly distributed between
321     * <code>lower</code> and <code>upper</code>, inclusive. This algorithm uses
322     * a secure random number generator.
323     *
324     * @param lower
325     *            the lower bound.
326     * @param upper
327     *            the upper bound.
328     * @return the random integer.
329     * @throws NumberIsTooLargeException if {@code lower >= upper}.
330     */
331    public long nextSecureLong(long lower, long upper) {
332        if (lower >= upper) {
333            throw new NumberIsTooLargeException(LocalizedFormats.LOWER_BOUND_NOT_BELOW_UPPER_BOUND,
334                                                lower, upper, false);
335        }
336        SecureRandom sec = getSecRan();
337        return lower + (long) (sec.nextDouble() * (upper - lower + 1));
338    }
339
340    /**
341     * {@inheritDoc}
342     * <p>
343     * <strong>Algorithm Description</strong>:
344     * <ul><li> For small means, uses simulation of a Poisson process
345     * using Uniform deviates, as described
346     * <a href="http://irmi.epfl.ch/cmos/Pmmi/interactive/rng7.htm"> here.</a>
347     * The Poisson process (and hence value returned) is bounded by 1000 * mean.</li>
348     *
349     * <li> For large means, uses the rejection algorithm described in <br/>
350     * Devroye, Luc. (1981).<i>The Computer Generation of Poisson Random Variables</i>
351     * <strong>Computing</strong> vol. 26 pp. 197-207.</li></ul></p>
352     *
353     * @param mean mean of the Poisson distribution.
354     * @return the random Poisson value.
355     * @throws NotStrictlyPositiveException if {@code mean <= 0}.
356     */
357    public long nextPoisson(double mean) {
358        if (mean <= 0) {
359            throw new NotStrictlyPositiveException(LocalizedFormats.MEAN, mean);
360        }
361
362        final RandomGenerator generator = getRan();
363
364        final double pivot = 40.0d;
365        if (mean < pivot) {
366            double p = FastMath.exp(-mean);
367            long n = 0;
368            double r = 1.0d;
369            double rnd = 1.0d;
370
371            while (n < 1000 * mean) {
372                rnd = generator.nextDouble();
373                r = r * rnd;
374                if (r >= p) {
375                    n++;
376                } else {
377                    return n;
378                }
379            }
380            return n;
381        } else {
382            final double lambda = FastMath.floor(mean);
383            final double lambdaFractional = mean - lambda;
384            final double logLambda = FastMath.log(lambda);
385            final double logLambdaFactorial = MathUtils.factorialLog((int) lambda);
386            final long y2 = lambdaFractional < Double.MIN_VALUE ? 0 : nextPoisson(lambdaFractional);
387            final double delta = FastMath.sqrt(lambda * FastMath.log(32 * lambda / FastMath.PI + 1));
388            final double halfDelta = delta / 2;
389            final double twolpd = 2 * lambda + delta;
390            final double a1 = FastMath.sqrt(FastMath.PI * twolpd) * FastMath.exp(1 / 8 * lambda);
391            final double a2 = (twolpd / delta) * FastMath.exp(-delta * (1 + delta) / twolpd);
392            final double aSum = a1 + a2 + 1;
393            final double p1 = a1 / aSum;
394            final double p2 = a2 / aSum;
395            final double c1 = 1 / (8 * lambda);
396
397            double x = 0;
398            double y = 0;
399            double v = 0;
400            int a = 0;
401            double t = 0;
402            double qr = 0;
403            double qa = 0;
404            for (;;) {
405                final double u = nextUniform(0.0, 1);
406                if (u <= p1) {
407                    final double n = nextGaussian(0d, 1d);
408                    x = n * FastMath.sqrt(lambda + halfDelta) - 0.5d;
409                    if (x > delta || x < -lambda) {
410                        continue;
411                    }
412                    y = x < 0 ? FastMath.floor(x) : FastMath.ceil(x);
413                    final double e = nextExponential(1d);
414                    v = -e - (n * n / 2) + c1;
415                } else {
416                    if (u > p1 + p2) {
417                        y = lambda;
418                        break;
419                    } else {
420                        x = delta + (twolpd / delta) * nextExponential(1d);
421                        y = FastMath.ceil(x);
422                        v = -nextExponential(1d) - delta * (x + 1) / twolpd;
423                    }
424                }
425                a = x < 0 ? 1 : 0;
426                t = y * (y + 1) / (2 * lambda);
427                if (v < -t && a == 0) {
428                    y = lambda + y;
429                    break;
430                }
431                qr = t * ((2 * y + 1) / (6 * lambda) - 1);
432                qa = qr - (t * t) / (3 * (lambda + a * (y + 1)));
433                if (v < qa) {
434                    y = lambda + y;
435                    break;
436                }
437                if (v > qr) {
438                    continue;
439                }
440                if (v < y * logLambda - MathUtils.factorialLog((int) (y + lambda)) + logLambdaFactorial) {
441                    y = lambda + y;
442                    break;
443                }
444            }
445            return y2 + (long) y;
446        }
447    }
448
449    /**
450     * Generate a random value from a Normal (a.k.a. Gaussian) distribution with
451     * the given mean, <code>mu</code> and the given standard deviation,
452     * <code>sigma</code>.
453     *
454     * @param mu
455     *            the mean of the distribution
456     * @param sigma
457     *            the standard deviation of the distribution
458     * @return the random Normal value
459     * @throws NotStrictlyPositiveException if {@code sigma <= 0}.
460     */
461    public double nextGaussian(double mu, double sigma) {
462        if (sigma <= 0) {
463            throw new NotStrictlyPositiveException(LocalizedFormats.STANDARD_DEVIATION, sigma);
464        }
465        return sigma * getRan().nextGaussian() + mu;
466    }
467
468    /**
469     * Returns a random value from an Exponential distribution with the given
470     * mean.
471     * <p>
472     * <strong>Algorithm Description</strong>: Uses the <a
473     * href="http://www.jesus.ox.ac.uk/~clifford/a5/chap1/node5.html"> Inversion
474     * Method</a> to generate exponentially distributed random values from
475     * uniform deviates.
476     * </p>
477     *
478     * @param mean the mean of the distribution
479     * @return the random Exponential value
480     * @throws NotStrictlyPositiveException if {@code mean <= 0}.
481     */
482    public double nextExponential(double mean) {
483        if (mean <= 0.0) {
484            throw new NotStrictlyPositiveException(LocalizedFormats.MEAN, mean);
485        }
486        final RandomGenerator generator = getRan();
487        double unif = generator.nextDouble();
488        while (unif == 0.0d) {
489            unif = generator.nextDouble();
490        }
491        return -mean * FastMath.log(unif);
492    }
493
494    /**
495     * {@inheritDoc}
496     * <p>
497     * <strong>Algorithm Description</strong>: scales the output of
498     * Random.nextDouble(), but rejects 0 values (i.e., will generate another
499     * random double if Random.nextDouble() returns 0). This is necessary to
500     * provide a symmetric output interval (both endpoints excluded).
501     * </p>
502     *
503     * @param lower
504     *            the lower bound.
505     * @param upper
506     *            the upper bound.
507     * @return a uniformly distributed random value from the interval (lower,
508     *         upper)
509     * @throws NumberIsTooLargeException if {@code lower >= upper}.
510     */
511    public double nextUniform(double lower, double upper) {
512        if (lower >= upper) {
513            throw new NumberIsTooLargeException(LocalizedFormats.LOWER_BOUND_NOT_BELOW_UPPER_BOUND,
514                                                lower, upper, false);
515        }
516        final RandomGenerator generator = getRan();
517
518        // ensure nextDouble() isn't 0.0
519        double u = generator.nextDouble();
520        while (u <= 0.0) {
521            u = generator.nextDouble();
522        }
523
524        return lower + u * (upper - lower);
525    }
526
527    /**
528     * Generates a random value from the {@link BetaDistributionImpl Beta Distribution}.
529     * This implementation uses {@link #nextInversionDeviate(ContinuousDistribution) inversion}
530     * to generate random values.
531     *
532     * @param alpha first distribution shape parameter
533     * @param beta second distribution shape parameter
534     * @return random value sampled from the beta(alpha, beta) distribution
535     * @throws MathException if an error occurs generating the random value
536     * @since 2.2
537     */
538    public double nextBeta(double alpha, double beta) throws MathException {
539        return nextInversionDeviate(new BetaDistributionImpl(alpha, beta));
540    }
541
542    /**
543     * Generates a random value from the {@link BinomialDistributionImpl Binomial Distribution}.
544     * This implementation uses {@link #nextInversionDeviate(ContinuousDistribution) inversion}
545     * to generate random values.
546     *
547     * @param numberOfTrials number of trials of the Binomial distribution
548     * @param probabilityOfSuccess probability of success of the Binomial distribution
549     * @return random value sampled from the Binomial(numberOfTrials, probabilityOfSuccess) distribution
550     * @throws MathException if an error occurs generating the random value
551     * @since 2.2
552     */
553    public int nextBinomial(int numberOfTrials, double probabilityOfSuccess) throws MathException {
554        return nextInversionDeviate(new BinomialDistributionImpl(numberOfTrials, probabilityOfSuccess));
555    }
556
557    /**
558     * Generates a random value from the {@link CauchyDistributionImpl Cauchy Distribution}.
559     * This implementation uses {@link #nextInversionDeviate(ContinuousDistribution) inversion}
560     * to generate random values.
561     *
562     * @param median the median of the Cauchy distribution
563     * @param scale the scale parameter of the Cauchy distribution
564     * @return random value sampled from the Cauchy(median, scale) distribution
565     * @throws MathException if an error occurs generating the random value
566     * @since 2.2
567     */
568    public double nextCauchy(double median, double scale) throws MathException {
569        return nextInversionDeviate(new CauchyDistributionImpl(median, scale));
570    }
571
572    /**
573     * Generates a random value from the {@link ChiSquaredDistributionImpl ChiSquare Distribution}.
574     * This implementation uses {@link #nextInversionDeviate(ContinuousDistribution) inversion}
575     * to generate random values.
576     *
577     * @param df the degrees of freedom of the ChiSquare distribution
578     * @return random value sampled from the ChiSquare(df) distribution
579     * @throws MathException if an error occurs generating the random value
580     * @since 2.2
581     */
582    public double nextChiSquare(double df) throws MathException {
583        return nextInversionDeviate(new ChiSquaredDistributionImpl(df));
584    }
585
586    /**
587     * Generates a random value from the {@link FDistributionImpl F Distribution}.
588     * This implementation uses {@link #nextInversionDeviate(ContinuousDistribution) inversion}
589     * to generate random values.
590     *
591     * @param numeratorDf the numerator degrees of freedom of the F distribution
592     * @param denominatorDf the denominator degrees of freedom of the F distribution
593     * @return random value sampled from the F(numeratorDf, denominatorDf) distribution
594     * @throws MathException if an error occurs generating the random value
595     * @since 2.2
596     */
597    public double nextF(double numeratorDf, double denominatorDf) throws MathException {
598        return nextInversionDeviate(new FDistributionImpl(numeratorDf, denominatorDf));
599    }
600
601    /**
602     * Generates a random value from the {@link GammaDistributionImpl Gamma Distribution}.
603     * This implementation uses {@link #nextInversionDeviate(ContinuousDistribution) inversion}
604     * to generate random values.
605     *
606     * @param shape the median of the Gamma distribution
607     * @param scale the scale parameter of the Gamma distribution
608     * @return random value sampled from the Gamma(shape, scale) distribution
609     * @throws MathException if an error occurs generating the random value
610     * @since 2.2
611     */
612    public double nextGamma(double shape, double scale) throws MathException {
613        return nextInversionDeviate(new GammaDistributionImpl(shape, scale));
614    }
615
616    /**
617     * Generates a random value from the {@link HypergeometricDistributionImpl Hypergeometric Distribution}.
618     * This implementation uses {@link #nextInversionDeviate(IntegerDistribution) inversion}
619     * to generate random values.
620     *
621     * @param populationSize the population size of the Hypergeometric distribution
622     * @param numberOfSuccesses number of successes in the population of the Hypergeometric distribution
623     * @param sampleSize the sample size of the Hypergeometric distribution
624     * @return random value sampled from the Hypergeometric(numberOfSuccesses, sampleSize) distribution
625     * @throws MathException if an error occurs generating the random value
626     * @since 2.2
627     */
628    public int nextHypergeometric(int populationSize, int numberOfSuccesses, int sampleSize) throws MathException {
629        return nextInversionDeviate(new HypergeometricDistributionImpl(populationSize, numberOfSuccesses, sampleSize));
630    }
631
632    /**
633     * Generates a random value from the {@link PascalDistributionImpl Pascal Distribution}.
634     * This implementation uses {@link #nextInversionDeviate(IntegerDistribution) inversion}
635     * to generate random values.
636     *
637     * @param r the number of successes of the Pascal distribution
638     * @param p the probability of success of the Pascal distribution
639     * @return random value sampled from the Pascal(r, p) distribution
640     * @throws MathException if an error occurs generating the random value
641     * @since 2.2
642     */
643    public int nextPascal(int r, double p) throws MathException {
644        return nextInversionDeviate(new PascalDistributionImpl(r, p));
645    }
646
647    /**
648     * Generates a random value from the {@link TDistributionImpl T Distribution}.
649     * This implementation uses {@link #nextInversionDeviate(ContinuousDistribution) inversion}
650     * to generate random values.
651     *
652     * @param df the degrees of freedom of the T distribution
653     * @return random value from the T(df) distribution
654     * @throws MathException if an error occurs generating the random value
655     * @since 2.2
656     */
657    public double nextT(double df) throws MathException {
658        return nextInversionDeviate(new TDistributionImpl(df));
659    }
660
661    /**
662     * Generates a random value from the {@link WeibullDistributionImpl Weibull Distribution}.
663     * This implementation uses {@link #nextInversionDeviate(ContinuousDistribution) inversion}
664     * to generate random values.
665     *
666     * @param shape the shape parameter of the Weibull distribution
667     * @param scale the scale parameter of the Weibull distribution
668     * @return random value sampled from the Weibull(shape, size) distribution
669     * @throws MathException if an error occurs generating the random value
670     * @since 2.2
671     */
672    public double nextWeibull(double shape, double scale) throws MathException {
673        return nextInversionDeviate(new WeibullDistributionImpl(shape, scale));
674    }
675
676    /**
677     * Generates a random value from the {@link ZipfDistributionImpl Zipf Distribution}.
678     * This implementation uses {@link #nextInversionDeviate(IntegerDistribution) inversion}
679     * to generate random values.
680     *
681     * @param numberOfElements the number of elements of the ZipfDistribution
682     * @param exponent the exponent of the ZipfDistribution
683     * @return random value sampled from the Zipf(numberOfElements, exponent) distribution
684     * @throws MathException if an error occurs generating the random value
685     * @since 2.2
686     */
687    public int nextZipf(int numberOfElements, double exponent) throws MathException {
688        return nextInversionDeviate(new ZipfDistributionImpl(numberOfElements, exponent));
689    }
690
691    /**
692     * Returns the RandomGenerator used to generate non-secure random data.
693     * <p>
694     * Creates and initializes a default generator if null.
695     * </p>
696     *
697     * @return the Random used to generate random data
698     * @since 1.1
699     */
700    private RandomGenerator getRan() {
701        if (rand == null) {
702            rand = new JDKRandomGenerator();
703            rand.setSeed(System.currentTimeMillis());
704        }
705        return rand;
706    }
707
708    /**
709     * Returns the SecureRandom used to generate secure random data.
710     * <p>
711     * Creates and initializes if null.
712     * </p>
713     *
714     * @return the SecureRandom used to generate secure random data
715     */
716    private SecureRandom getSecRan() {
717        if (secRand == null) {
718            secRand = new SecureRandom();
719            secRand.setSeed(System.currentTimeMillis());
720        }
721        return secRand;
722    }
723
724    /**
725     * Reseeds the random number generator with the supplied seed.
726     * <p>
727     * Will create and initialize if null.
728     * </p>
729     *
730     * @param seed
731     *            the seed value to use
732     */
733    public void reSeed(long seed) {
734        if (rand == null) {
735            rand = new JDKRandomGenerator();
736        }
737        rand.setSeed(seed);
738    }
739
740    /**
741     * Reseeds the secure random number generator with the current time in
742     * milliseconds.
743     * <p>
744     * Will create and initialize if null.
745     * </p>
746     */
747    public void reSeedSecure() {
748        if (secRand == null) {
749            secRand = new SecureRandom();
750        }
751        secRand.setSeed(System.currentTimeMillis());
752    }
753
754    /**
755     * Reseeds the secure random number generator with the supplied seed.
756     * <p>
757     * Will create and initialize if null.
758     * </p>
759     *
760     * @param seed
761     *            the seed value to use
762     */
763    public void reSeedSecure(long seed) {
764        if (secRand == null) {
765            secRand = new SecureRandom();
766        }
767        secRand.setSeed(seed);
768    }
769
770    /**
771     * Reseeds the random number generator with the current time in
772     * milliseconds.
773     */
774    public void reSeed() {
775        if (rand == null) {
776            rand = new JDKRandomGenerator();
777        }
778        rand.setSeed(System.currentTimeMillis());
779    }
780
781    /**
782     * Sets the PRNG algorithm for the underlying SecureRandom instance using
783     * the Security Provider API. The Security Provider API is defined in <a
784     * href =
785     * "http://java.sun.com/j2se/1.3/docs/guide/security/CryptoSpec.html#AppA">
786     * Java Cryptography Architecture API Specification & Reference.</a>
787     * <p>
788     * <strong>USAGE NOTE:</strong> This method carries <i>significant</i>
789     * overhead and may take several seconds to execute.
790     * </p>
791     *
792     * @param algorithm
793     *            the name of the PRNG algorithm
794     * @param provider
795     *            the name of the provider
796     * @throws NoSuchAlgorithmException
797     *             if the specified algorithm is not available
798     * @throws NoSuchProviderException
799     *             if the specified provider is not installed
800     */
801    public void setSecureAlgorithm(String algorithm, String provider)
802            throws NoSuchAlgorithmException, NoSuchProviderException {
803        secRand = SecureRandom.getInstance(algorithm, provider);
804    }
805
806    /**
807     * Generates an integer array of length <code>k</code> whose entries are
808     * selected randomly, without repetition, from the integers
809     * <code>0 through n-1</code> (inclusive).
810     * <p>
811     * Generated arrays represent permutations of <code>n</code> taken
812     * <code>k</code> at a time.
813     * </p>
814     * <p>
815     * <strong>Preconditions:</strong>
816     * <ul>
817     * <li> <code>k <= n</code></li>
818     * <li> <code>n > 0</code></li>
819     * </ul>
820     * If the preconditions are not met, an IllegalArgumentException is thrown.
821     * </p>
822     * <p>
823     * Uses a 2-cycle permutation shuffle. The shuffling process is described <a
824     * href="http://www.maths.abdn.ac.uk/~igc/tch/mx4002/notes/node83.html">
825     * here</a>.
826     * </p>
827     *
828     * @param n
829     *            domain of the permutation (must be positive)
830     * @param k
831     *            size of the permutation (must satisfy 0 < k <= n).
832     * @return the random permutation as an int array
833     * @throws NumberIsTooLargeException if {@code k > n}.
834     * @throws NotStrictlyPositiveException if {@code k <= 0}.
835     */
836    public int[] nextPermutation(int n, int k) {
837        if (k > n) {
838            throw new NumberIsTooLargeException(LocalizedFormats.PERMUTATION_EXCEEDS_N,
839                                                k, n, true);
840        }
841        if (k == 0) {
842            throw new NotStrictlyPositiveException(LocalizedFormats.PERMUTATION_SIZE,
843                                                   k);
844        }
845
846        int[] index = getNatural(n);
847        shuffle(index, n - k);
848        int[] result = new int[k];
849        for (int i = 0; i < k; i++) {
850            result[i] = index[n - i - 1];
851        }
852
853        return result;
854    }
855
856    /**
857     * Uses a 2-cycle permutation shuffle to generate a random permutation.
858     * <strong>Algorithm Description</strong>: Uses a 2-cycle permutation
859     * shuffle to generate a random permutation of <code>c.size()</code> and
860     * then returns the elements whose indexes correspond to the elements of the
861     * generated permutation. This technique is described, and proven to
862     * generate random samples, <a
863     * href="http://www.maths.abdn.ac.uk/~igc/tch/mx4002/notes/node83.html">
864     * here</a>
865     *
866     * @param c
867     *            Collection to sample from.
868     * @param k
869     *            sample size.
870     * @return the random sample.
871     * @throws NumberIsTooLargeException if {@code k > c.size()}.
872     * @throws NotStrictlyPositiveException if {@code k <= 0}.
873     */
874    public Object[] nextSample(Collection<?> c, int k) {
875        int len = c.size();
876        if (k > len) {
877            throw new NumberIsTooLargeException(LocalizedFormats.SAMPLE_SIZE_EXCEEDS_COLLECTION_SIZE,
878                                                k, len, true);
879        }
880        if (k <= 0) {
881            throw new NotStrictlyPositiveException(LocalizedFormats.NUMBER_OF_SAMPLES, k);
882        }
883
884        Object[] objects = c.toArray();
885        int[] index = nextPermutation(len, k);
886        Object[] result = new Object[k];
887        for (int i = 0; i < k; i++) {
888            result[i] = objects[index[i]];
889        }
890        return result;
891    }
892
893    /**
894     * Generate a random deviate from the given distribution using the
895     * <a href="http://en.wikipedia.org/wiki/Inverse_transform_sampling"> inversion method.</a>
896     *
897     * @param distribution Continuous distribution to generate a random value from
898     * @return a random value sampled from the given distribution
899     * @throws MathException if an error occurs computing the inverse cumulative distribution function
900     * @since 2.2
901     */
902    public double nextInversionDeviate(ContinuousDistribution distribution) throws MathException {
903        return distribution.inverseCumulativeProbability(nextUniform(0, 1));
904
905    }
906
907    /**
908     * Generate a random deviate from the given distribution using the
909     * <a href="http://en.wikipedia.org/wiki/Inverse_transform_sampling"> inversion method.</a>
910     *
911     * @param distribution Integer distribution to generate a random value from
912     * @return a random value sampled from the given distribution
913     * @throws MathException if an error occurs computing the inverse cumulative distribution function
914     * @since 2.2
915     */
916    public int nextInversionDeviate(IntegerDistribution distribution) throws MathException {
917        final double target = nextUniform(0, 1);
918        final int glb = distribution.inverseCumulativeProbability(target);
919        if (distribution.cumulativeProbability(glb) == 1.0d) { // No mass above
920            return glb;
921        } else {
922            return glb + 1;
923        }
924    }
925
926    // ------------------------Private methods----------------------------------
927
928    /**
929     * Uses a 2-cycle permutation shuffle to randomly re-order the last elements
930     * of list.
931     *
932     * @param list
933     *            list to be shuffled
934     * @param end
935     *            element past which shuffling begins
936     */
937    private void shuffle(int[] list, int end) {
938        int target = 0;
939        for (int i = list.length - 1; i >= end; i--) {
940            if (i == 0) {
941                target = 0;
942            } else {
943                target = nextInt(0, i);
944            }
945            int temp = list[target];
946            list[target] = list[i];
947            list[i] = temp;
948        }
949    }
950
951    /**
952     * Returns an array representing n.
953     *
954     * @param n
955     *            the natural number to represent
956     * @return array with entries = elements of n
957     */
958    private int[] getNatural(int n) {
959        int[] natural = new int[n];
960        for (int i = 0; i < n; i++) {
961            natural[i] = i;
962        }
963        return natural;
964    }
965
966}
967